DICE Pre-Test

How Much Is Enough? Exploring Frequency Capping in Social Media Advertising

Authors
Affiliations

Hauke Roggenkamp

Institute of Behavioral Science and Technology, University of St. Gallen

Johannes Boegershausen

Rotterdam School of Management, Erasmus University

Christian Hildebrand

Institute of Behavioral Science and Technology, University of St. Gallen

Published

Saturday Jun 22, 2024, 08:01 GMT+2

The identifiable victim effect is the tendency show more empathy to an identifiable individual over a group of unidentified victims who are described using numerical statistics (see, e.g., Jenni and Loewenstein 1997; Small and Loewenstein 2003; Maier, Wong, and Feldman 2023). One significant study in this domain is by Small, Loewenstein, and Slovic (2007) demonstrates that donors are more likely to contribute to charitable causes when presented with specific stories about individuals, or “identified victims”, rather than abstract statistics about large groups. They found that personal stories evoke stronger emotional connections.

We use DICE to create social media feeds where organic posts and advertisements compete for the participant’s attention to study whether identified victims (compared to abstract statistics) cause social media advertisements to more effectively “cut through the content clutter” (Ordenes et al. 2019) and drive ad recall for the charities posting these ads.

A key managerial question for any organization, but particularly for organizations with limited budgets and resources, like charities, is to understand how frequently they need to show their advertising to effectively cut through the clutter and generate positive outcomes from the digital ads. Particularly effective ads will likely need less exposure to users to produce positive outcomes like brand recall while avoiding negative reactions such as ad fatigue.

A key lever in online advertising is Frequency Capping (FC), which limits how often a specific advertisement is shown to the same user within a set period. It is essential for preventing ad fatigue (see, e.g., Braun and Moe 2013; Silberstein, Shoham, and Klein 2023), where users become desensitized to an ad due to excessive exposure, which can lead to diminished engagement rates and a negative user experience. In addition to these indirect costs, high caps also have direct costs as the marginal effect of ad exposure can be assumed to be diminishing. By setting an optimal cap, marketers aim to balance both direct and indirect costs as well as benefits associated to an ad’s visibility.

Integrating these concepts, this study employs a 2 (identified victim vs. abstract numbers) by 4 (number of ad impressions) between-subjects design to investigate whether and how the identified victim effect drives ad recall over the course of varying degrees of ad penetration

Experimental Design

On a high level, we create different feeds that contain both organic and sponsored posts as illustrated in Figure 1, where the first column on the left illustrates a feed that contains five sets of organic posts (grey boxes), three filler ads (yellow, green, and blue boxes) as well as one focal ad, which is either v.1 (showing an identified victim) or v.2 (showing abstract numbers). Moving from left to the right, the figure shows how the filler ads are successively replaced by the focal ad, essentially manipulating the amount of (focal-) ad impressions per feed.

Figure 1: 2 (identified victim vs. abstract numbers) by 4 (number of focal ad impressions) between-subjects design

Given a frequency cap \(\{1, 2, 3, 4\}\), we do neither vary the organic posts nor the filler ads between conditions. However, we do manipulate the focal ad (identified victim vs. abstract numbers). In addition, the order in which the ads are displayed is randomized between subjects too. Hence, some participants may encounter the ads in a sequence like Ad A > Ad C > Ad B > Focal Ad, whereas other see a sequence like Focal Ad > Ad A > Ad C > Ad B or any other variation. The “slots” in which an ad can be displayed remain constant.1

Stimuli

The two variations of the focal ad creative are displayed in Figure 2.

Figure 2: Variations of Focal Ad

(a) Identified Victim Condition

(b) Abstract Condition

In addition, we vary the ad copy:

  • Identified Victim Condition: »I just want to go home.« whispers João, 7, displaced by the floods in Brazil. His story is one of many, but today, you can make a difference for him. Help João and others like him find safety. Donate now.

    #HelpJoao #BrazilFloodRelief
  • Abstract Condition: Thousands have been displaced by the floods in Brazil and want to go home. Today, you can make a difference for the those in need and help them find safety. Donate now.

    #HelpBrazil #BrazilFloodRelief

You can find a feed displaying the abstract ad here.

Before participants browse the feed, we brief them as follows:

Instructions: Welcome to our study covering a typical social media feed that focuses on the topic of Brazil. Please interact with the feed as you normally would on social media. Feel free to like, comment, or simply read through the posts according to your preference. Once you reach the end of the feed, you’ll find a button to proceed with the study. Click this button to move on to a series of short questions about your experience with the feed.

Pre-Test

We run a 2-cell (identified victim vs. abstract numbers) between-subjects design with two ad exposures as a pre-test (i.e., only the second column of Figure 1).

We pre-test the experiment as we implemented new features in the DICE software, that measure the height in which each post is resolved on a participant’s device, for instance. In addition, we measure whether ads are clicked. Most importantly, we also changed the mechanisms randomizing the conditions as well as potential randomizations of the order in which posts of a feed are displayed.

Primary Analysis

After participants browse the social media feed, they are redirected to a Qualtrics survey that starts with basic demographic questions. Subsequently, they answer unaided and aided recall questions to indicate whether they remember seeing a unicef ad.

Our primary interest lies in the identified victim effect on these recall measures. We expect higher recall on ads with identifiable victims, which is why we consider a one sided test.

Population

We will recruit participants from Prolific who haven’t participated in other DICE studies before and meet the following additional criteria:

  • Approval Rate >= 99%
  • First Language == ‘English’
  • Location == ‘USA’

Sample Size

We recruit 100 participants in a first pilot. To this end, we create a database containing 200 rows. The corresponding session-code is mkkum3r5.

Exclusion Criteria

We will only consider complete observations, that is, data from participants who browsed through the feed, answered the qualtrics survey and who were redirected to Prolific with a functional completion code.

Because we gather process data, such as dwell time, we have tools to assess the data quality (Cuskley and Sulik) – at least during the exposure to the social media feed. If these data reveal inattentive participants, for instance, we may exclude them too but label the resulting analyses as exploratory.

Prior Data Collection

We did not collect any data before.

References

Braun, Michael, and Wendy W. Moe. 2013. “Online Display Advertising: Modeling the Effects of Multiple Creatives and Individual Impression Histories.” Marketing Science 32 (5): 753–67. https://doi.org/10.1287/mksc.2013.0802.
Cuskley, Christine, and Justin Sulik. “The Burden for High-Quality Online Data Collection Lies with Researchers, Not Recruitment Platforms.” Perspectives on Psychological Science 0 (0): 17456916241242734. https://doi.org/10.1177/17456916241242734.
Jenni, Karen E., and George Loewenstein. 1997. “Explaining the Identifiable Victim Effect.” Journal of Risk and Uncertainty 14 (3): 235–57. https://doi.org/10.1023/A:1007740225484.
Maier, Maximilian, Yik Chun Wong, and Gilad Feldman. 2023. “Revisiting and Rethinking the Identifiable Victim Effect: Replication and Extension of Small, Loewenstein, and Slovic (2007).” Collabra: Psychology 9 (1): 90203. https://doi.org/10.1525/collabra.90203.
Ordenes, Francisco Villarroel, Dhruv Grewal, Stephan Ludwig, Ko De Ruyter, Dominik Mahr, and Martin Wetzels. 2019. “Cutting Through Content Clutter: How Speech and Image Acts Drive Consumer Sharing of Social Media Brand Messages.” Journal of Consumer Research 45 (5): 988–1012. https://doi.org/10.1093/jcr/ucy032.
Silberstein, Natalia, Or Shoham, and Assaf Klein. 2023. “Combating Ad Fatigue via Frequency-Recency Features in Online Advertising Systems.” In Proceedings of the 32nd ACM International Conference on Information and Knowledge Management, 4822–28. CIKM ’23. New York, NY, USA: Association for Computing Machinery. https://doi.org/10.1145/3583780.3615461.
Small, Deborah A., and George Loewenstein. 2003. “Helping a Victim or Helping the Victim: Altruism and Identifiability.” Journal of Risk and Uncertainty 26 (1): 5–16. https://doi.org/10.1023/A:1022299422219.
Small, Deborah A., George Loewenstein, and Paul Slovic. 2007. “Sympathy and Callousness: The Impact of Deliberative Thought on Donations to Identifiable and Statistical Victims.” Organizational Behavior and Human Decision Processes 102 (2): 143–53. https://doi.org/10.1016/j.obhdp.2006.01.005.

Footnotes

  1. We expose participants to ads in 6th, 21st, 36st, and 51st position of the feed.↩︎